def rdfStarLinkGenerator_fromCSV(link_predicate: str, result_batch, offset=0):
errors = ""
vars_size = 0
buffer = Buffer()
vars_dic = defaultdict(int)
for count, row in enumerate(result_batch):
if True:
# THE FIRST LINE IS ASSUMED TO BE THE HEADER
if count > 0 and len(row) > 1:
# GET THE SOURCE AND TARGET URIS
src_data, trg_data = row[0], row[1]
# GENERATION OF THE LINK
if src_data and trg_data:
# The RDFStar subject
buffer.write(F"{space}### LINK Nbr: {count + offset}\n"
F"{space}<<<{src_data}> {link_predicate} <{trg_data}>>>\n"
if len(vars_dic) > 0
else F"{space}<{src_data}> {link_predicate} <{trg_data}> .\n")
# ANNOTATION OF THE LINK
# ll_val:has-link-validation "not_validated" .
for counter, (predicate, index) in enumerate(vars_dic.items()):
end = ".\n" if counter == vars_size - 1 else ";"
# APPENDING THE CLUSTER SIZE
# if clusters and predicate == VoidPlus.cluster_ID_ttl and int(row[index]) in clusters:
# buffer.write(F"{space * 2}{VoidPlus.cluster_size_ttl:{Vars.PRED_SIZE}}"
# F"{Literal(clusters[int(row[index])]['size']).n3(MANAGER)} ;\n")
# APPENDING THE VALIDATION FLAG
# if predicate == VoidPlus.has_validation_flag_ttl:
# triple_value = validate.get_resource[row[index]]
# APPENDING THE VALIDATION FLAG RESOURCE
if predicate == VoidPlus.has_validation_ttl:
small = src_data if src_data < trg_data else trg_data
big = trg_data if small == src_data else src_data
key = Grl.deterministicHash(F"{small}{big}{link_predicate}")
triple_value = Rsc.validation_ttl(key)
# buffer.write(F"{space * 2}{VoidPlus.has_validation_ttl:{Vars.PRED_SIZE}}{triple_value} {end}\n")
# APPENDING THE CLUSTER ID AS A RESOURCE
elif predicate == VoidPlus.cluster_ID_ttl:
cluster_id = int(row[index])
triple_value = Rsc.cluster_ttl(cluster_id)
# clusters[cluster_id]['item'].extend([src_data, trg_data])
# APPENDING ANYTHING ELSE
else:
triple_value = Literal(round(float(row[index]), 5)).n3(MANAGER) \
if Grl.isDecimalLike(row[index]) \
else Literal(row[index]).n3(MANAGER)
buffer.write(F"{space * 2}{predicate:{Vars.PRED_SIZE}}{triple_value} {end}\n")
# buffer.write(F"{space * 2}{predicate:{Vars.PRED_SIZE}}"
# F"{validate.get_resource[row[index]] if not Grl.isDecimalLike(row[index]) else round(float(row[index]), 5)} {end}\n")
yield buffer.getvalue()
clearBuffer(buffer)
else:
# THE CSV HEADER
# Star at position
for column in range(2, len(row)):
if row[column] in CSV_HEADERS:
vars_dic[CSV_HEADERS[row[column]]] = column
vars_size += 1
def standardLinkGenerator2(link_predicate: str, result_batch, namespace, clusters=None, offset=0):
"""
:param offset : an integer to increment the counting of tghe links
:param link_predicate : a turtle representation of a URI (e.i: owl:sameAs).
:param namespace : a dictionary for namespace
:param result_batch : an iterable object with link results.
:param clusters : a dictionary proving the size of the clusters links.
:return : Yields a string as set of triples.
"""
errors = ""
vars_size = 0
buffer = Buffer()
vars_dic = defaultdict(int)
for count, row in enumerate(result_batch):
try:
# THE FIRST LINE IS ASSUMED TO BE THE HEADER
if count > 0 and len(row) > 1:
# GET THE SOURCE AND TARGET URIS
src_data, trg_data, predicate = uri2ttl(row[0], namespace)["short"], \
uri2ttl(row[1], namespace)["short"], \
uri2ttl(link_predicate, namespace)["short"]
print(src_data)
# GENERATION OF THE LINK
if src_data and trg_data:
# The RDFStar subject
buffer.write(F"\n{space}### LINK Nbr: {count + offset}\n"
F"{space}{src_data} {Rsc.ga_resource_ttl(predicate)} {trg_data} .\n")
# STANDARD REIFICATION
link = F"{space}{src_data} {Rsc.ga_resource_ttl(predicate)} {trg_data} .\n"
code = Rsc.ga_resource_ttl(F"Reification-{Grl.deterministicHash(link)}")
buffer.write(F"\n{space}### STANDARD REIFICATION Nbr: {count}"
F"\n{space}{code}\n"
F"{space}{preVal('a', 'rdf:Statement')}"
F"{space}{preVal('rdf:predicate', predicate)}"
F"{space}{preVal('rdf:subject', F'{src_data}')}"
F"{space}{preVal('rdf:object', F'{trg_data}')}")
# ANNOTATION OF THE LINK USING THE REIFIED CODE
for counter, (predicate, index) in enumerate(vars_dic.items()):
end = ".\n" if counter == vars_size - 1 else ";"
# APPENDING THE CLUSTER SIZE
if clusters and predicate == VoidPlus.cluster_ID_ttl and row[index] in clusters:
buffer.write(F"{space * 2}{VoidPlus.cluster_size_ttl:{Vars.PRED_SIZE}}"
F"{Literal(clusters[row[index]]).n3(MANAGER)} ;\n")
# APPENDING THE VALIDATION FLAG
if predicate == VoidPlus.has_validation_status_ttl:
triple_value = validate.get_resource[row[index]]
# APPENDING DING ANYTHING ELSE
else:
triple_value = Literal(round(float(row[index]), 5)).n3(MANAGER) \
if Grl.isDecimalLike(row[index]) \
else Literal(row[index]).n3(MANAGER)
buffer.write(F"{space * 2}{predicate:{Vars.PRED_SIZE}}{triple_value} {end}\n")
yield buffer.getvalue()
clearBuffer(buffer)
else:
# THE CSV HEADER
# Star at position
# MAPPING THE CSV HEADERS
for column in range(2, len(row)):
header = row
if row[column] in CSV_HEADERS:
vars_dic[CSV_HEADERS[row[column]]] = column
vars_size += 1
except Exception as err:
errors += F">>>> [ERROR FROM csv_2_linkset] {row}, {err}"
print(errors)
def rdfStarLinkGenerator(mappings: dict, link_predicate: str, result_batch, offset=0):
errors = ""
buffer = Buffer()
def ns_modification(uri):
for ns in mappings:
if uri.startswith(ns):
uri = uri.replace(ns, F"{mappings[ns]}:")
break
if uri.__contains__("://"):
uri = F"<{uri}>"
return uri
for count, link in enumerate(result_batch):
try:
# GET THE SOURCE AND TARGET URIS
src_data, trg_data = ns_modification(link['source']), ns_modification(link['target'])
# GENERATION OF THE LINK
if src_data and trg_data:
# The RDFStar subject
buffer.write(F"{space}### LINK Nbr: {count + offset}\n"
F"{space}<<{src_data} {link_predicate} {trg_data}>>\n")
# ANNOTATION OF THE LINK
# ll_val:has-link-validation "not_validated" .
for counter, (feature, value) in enumerate(link.items()):
end = ".\n" if counter == len(link) - 1 else ";"
current_property = JSON_HEADERS.get(feature, None)
if current_property:
# APPENDING THE VALIDATION FLAG RESOURCE
if current_property == VoidPlus.has_validation_ttl:
small = link['source'] if link['source'] < link['target'] else link['target']
big = link['target'] if small == link['source'] else link['source']
key = Grl.deterministicHash(F"{small}{big}{link_predicate}")
triple_value = Rsc.validation_ttl(key) if key is not None else key
# NOT APPENDING THE CLUSTER INT ID
elif current_property == VoidPlus.cluster_ID_ttl:
triple_value = Rsc.cluster_ttl(value) if value is not None else value
# APPENDING ANYTHING ELSE
else:
if current_property == VoidPlus.cluster_Int_ID_ttl:
triple_value = None
elif value is not None:
triple_value = Literal(round(float(value), 5)).n3(MANAGER) \
if Grl.isDecimalLike(value) \
else Literal(value).n3(MANAGER)
else:
triple_value = value
if triple_value is not None:
buffer.write(F"{space * 2}{current_property:{Vars.PRED_SIZE}}{triple_value} {end}\n")
yield buffer.getvalue()
clearBuffer(buffer)
except Exception as err:
errors += F">>>> [ERROR FROM AnnotatedLinkset_Generic/rdfStarLinkGenerator] {link}, {err}"
def standardLinkGenerator(mappings: dict, link_predicate: str, result_batch, offset=0):
"""
:param mappings : dictionary of namespaces as keys and prefixes ad values.
:param offset : an integer to increment the counting of tghe links
:param link_predicate : a turtle representation of a URI (e.i: owl:sameAs).
:param result_batch : an iterable object with link results.
:param clusters : a dictionary proving the size of the clusters links.
:return : Yields a string as set of triples.
"""
buffer = Buffer()
errors = ""
def ns_modification(uri):
for ns in mappings:
if uri.startswith(ns):
uri = uri.replace(ns, F"{mappings[ns]}:")
break
if uri.__contains__("://"):
uri = F"<{uri}>"
return uri
for count, link in enumerate(result_batch):
if True:
# GET THE SOURCE AND TARGET URIS
# src_data, trg_data = link['source'], link['target']
src_data, trg_data = ns_modification(link['source']), ns_modification(link['target'])
# GENERATION OF THE LINK
if src_data and trg_data:
# The RDFStar subject
buffer.write(F"\n{space}### LINK Nbr: {count + offset}\n"
F"{space}{src_data} {Rsc.ga_resource_ttl(link_predicate)} {trg_data} .\n")
# STANDARD REIFICATION
reification = F"{space}{src_data} {Rsc.ga_resource_ttl(link_predicate)} {trg_data} .\n"
code = Rsc.ga_resource_ttl(F"Reification-{Grl.deterministicHash(reification)}")
buffer.write(F"\n{space}### STANDARD REIFICATION Nbr: {count}"
F"\n{space}{code}\n"
F"{space}{preVal('a', 'rdf:Statement')}"
F"{space}{preVal('rdf:predicate', link_predicate)}"
F"{space}{preVal('rdf:subject', F'{src_data}')}"
F"{space}{preVal('rdf:object', F'{trg_data}')}")
# ANNOTATION OF THE LINK USING THE REIFIED CODE
for counter, (feature, value) in enumerate(link.items()):
end = ".\n" if counter == len(link) - 1 else ";"
cur_predicate = JSON_HEADERS.get(feature, None)
if cur_predicate:
# APPENDING THE VALIDATION FLAG RESOURCE
if cur_predicate == VoidPlus.has_validation_ttl:
small = link['source'] if link['source'] < link['target'] else link['target']
big = link['target'] if small == link['source'] else link['source']
# print(F"{small} {big} {link_predicate}")
key = Grl.deterministicHash(F"{small}{big}{link_predicate}")
triple_value = Rsc.validation_ttl(key) if key is not None else key
# APPENDING THE CLUSTER ID AS A RESOURCE
elif cur_predicate == VoidPlus.cluster_ID_ttl:
triple_value = Rsc.cluster_ttl(value) if value is not None else value
# triple_value = None
elif cur_predicate == VoidPlus.network_ID_ttl:
print("++++++++++++++++++>>>>>>>>>>")
triple_value = Literal(value).n3(MANAGER) if value is not None else value
# APPENDING ANYTHING ELSE
else:
if cur_predicate == VoidPlus.cluster_Int_ID_ttl:
triple_value = None
elif value is not None:
triple_value = Literal(round(float(value), 5)).n3(MANAGER) \
if Grl.isDecimalLike(value) \
else Literal(value).n3(MANAGER)
else:
triple_value = None
if triple_value is not None:
buffer.write(F"{space * 2}{cur_predicate:{Vars.PRED_SIZE}}{triple_value} {end}\n")
yield buffer.getvalue()
clearBuffer(buffer)
def standardLinkGenerator_fromCSV(link_predicate: str, result_batch, offset=0):
"""
:param offset : an integer to increment the counting of tghe links
:param link_predicate : a turtle representation of a URI (e.i: owl:sameAs).
:param result_batch : an iterable object with link results.
:param clusters : a dictionary proving the size of the clusters links.
:return : Yields a string as set of triples.
"""
errors = ""
vars_size = 0
buffer = Buffer()
vars_dic = defaultdict(int)
# print(clusters)
for count, row in enumerate(result_batch):
try:
# THE FIRST LINE IS ASSUMED TO BE THE HEADER
if count > 0 and len(row) > 1:
# GET THE SOURCE AND TARGET URIS
src_data, trg_data = row[0], row[1]
# GENERATION OF THE LINK
if src_data and trg_data:
# The RDFStar subject
buffer.write(F"\n{space}### LINK Nbr: {count + offset}\n"
F"{space}<{src_data}> {Rsc.ga_resource_ttl(link_predicate)} <{trg_data}> .\n")
# STANDARD REIFICATION
link = F"{space}{src_data} {Rsc.ga_resource_ttl(link_predicate)} {trg_data} .\n"
code = Rsc.ga_resource_ttl(F"Reification-{Grl.deterministicHash(link)}")
buffer.write(F"\n{space}### STANDARD REIFICATION Nbr: {count}"
F"\n{space}{code}\n"
F"{space}{preVal('a', 'rdf:Statement')}"
F"{space}{preVal('rdf:predicate', link_predicate)}"
F"{space}{preVal('rdf:subject', F'<{src_data}>')}"
F"{space}{preVal('rdf:object', F'<{trg_data}>')}")
# ANNOTATION OF THE LINK USING THE REIFIED CODE
for counter, (predicate, index) in enumerate(vars_dic.items()):
end = ".\n" if counter == vars_size - 1 else ";"
# APPENDING THE CLUSTER SIZE
# if clusters and predicate == VoidPlus.cluster_ID_ttl and int(row[index]) in clusters:
# buffer.write(F"{space * 2}{VoidPlus.cluster_size_ttl:{Vars.PRED_SIZE}}"
# F"{Literal(clusters[int(row[index])]['size']).n3(MANAGER)} ;\n")
# APPENDING THE VALIDATION FLAG
# if predicate == VoidPlus.has_validation_flag_ttl:
# triple_value = validate.get_resource[row[index]]
# APPENDING THE VALIDATION FLAG RESOURCE
if predicate == VoidPlus.has_validation_ttl:
small = src_data if src_data < trg_data else trg_data
big = trg_data if small == src_data else src_data
key = Grl.deterministicHash(F"{small}{big}{link_predicate}")
triple_value = Rsc.validation_ttl(key)
# buffer.write(F"{space * 2}{VoidPlus.has_validation_ttl:{Vars.PRED_SIZE}}{triple_value} {end}\n")
# APPENDING THE CLUSTER ID AS A RESOURCE
elif predicate == VoidPlus.cluster_ID_ttl:
cluster_id = int(row[index])
triple_value = Rsc.cluster_ttl(cluster_id)
# clusters[cluster_id]['item'].extend([src_data, trg_data])
# APPENDING ANYTHING ELSE
else:
triple_value = Literal(round(float(row[index]), 5)).n3(MANAGER) \
if Grl.isDecimalLike(row[index]) \
else Literal(row[index]).n3(MANAGER)
buffer.write(F"{space * 2}{predicate:{Vars.PRED_SIZE}}{triple_value} {end}\n")
yield buffer.getvalue()
clearBuffer(buffer)
else:
# THE CSV HEADER
# Star at position
# MAPPING THE CSV HEADERS
row_header = row
# print(header, len(header))
for column in range(2, len(row_header)):
if row[column] in CSV_HEADERS:
vars_dic[CSV_HEADERS[row_header[column]]] = column
# print('--->', CSV_HEADERS[header[column]], header[column], column)
vars_size += 1
except Exception as err:
errors += F">>>> [ERROR FROM AnnotatedLinkset_Generic/standardLinkGenerator] \n\t{row} \n\t{err}"
print(errors)